Maths (Probabilities) Puzzle.
Maths (Probabilities) Puzzle.
Author
Discussion

him_over_there

Original Poster:

970 posts

223 months

Tuesday 19th January 2010
quotequote all
Got a maths puzzle here I'm trying to solve, but I'm hopeless with probabilities.

Two players are playing a game with a coin.The coin is unbiased.

The game is turn based.

Player 1 - Tosses the coin. If it comes up heads Player 1 scores 1 point. If it comes up tails Player 1 scores 0 points.


Player 2 - Chooses an integer T and tosses the coin T times. If each toss is a head player 2 scores 2^(T-1) points. Otherwise player 2 scores 0.

Player 1 goes first.

Winner is first to 100 points.

Each time Player 2 takes a turn they choose a value for T which maximises their probability of winning the game.

Solving the chance of Player 2 winning seems simple if I knew how to determine each turn the value Player 2 should pick which gives the highest probability of winning. This is what I need help on smile

Anyone ?

ClareC

32 posts

237 months

Tuesday 19th January 2010
quotequote all
Doesn't matter, whatever value of T is choosen the "expected" return on each go would be half a point.

i.e. T=1, 50% chance of scoring 1, 50% chance of scoring 0.
T=2 25% chance of scoring 2, 75% chance of scoring 0
T=3 12.5% chance of scoring 4, 87.5% chance of scoring 0.

etc.

him_over_there

Original Poster:

970 posts

223 months

Tuesday 19th January 2010
quotequote all
That suggests that Player 2 and Player 1 both have the same chance of winning, both 50/50 if Player 2 is choosing T irrespective of Player 1's score.

Yet, the probability of P2 winning is required at an accuracy of 8 decimal places so their must be something I'm missing.

Size Nine Elm

5,167 posts

301 months

Tuesday 19th January 2010
quotequote all
Interesting...

Player 1 has obviously no bias, each turn can score 0 or 1 with 50% probability.
Player 2 can score an average of 0.5 each turn, same as player 1, regardless of T

So using a random value of T for player 2 will not bias the probability, and player 1 will win fractionally more than half the games, only because he goes first.

So player 2 must bias T depending on the actual scores at any time...

Probably worth working out probabilities for being player 2's turn with the scores at 99-99, then 99-98, 99-97, 98-98, and see if you can spot a pattern.

G_T

16,163 posts

207 months

Tuesday 19th January 2010
quotequote all
They were on the ground floor.


V8mate

45,899 posts

206 months

Tuesday 19th January 2010
quotequote all
G_T said:
They were on the ground floor.
Sure?

They were definitely both tossers.

sonic_2k_uk

4,008 posts

224 months

Tuesday 19th January 2010
quotequote all
P1 has a probability of 0.5 each turn to gain 1 point, so would take 200 turns to win.

If P2 sets T=8 each turn then they only need to hit it once to win (2^(8-1) = 128).

The probability of hitting 8 heads in a row is 0.5^8 = 0.00390625, thus the probability of them NOT hitting it each turn is 1-0.00390625 = 0.99609375.

The probability of not hitting it 200 times in a row is 0.99609375^200 = 0.45713347439222973069592389903499

V8mate

45,899 posts

206 months

Tuesday 19th January 2010
quotequote all
JDMFanYo said:
If player 2 tosses the coin 1 million times, the probability is that they will be first to get 100 points, right?

Edited by JDMFanYo on Tuesday 19th January 14:02
200 tosses should cover it.

him_over_there

Original Poster:

970 posts

223 months

Tuesday 19th January 2010
quotequote all
JDMFanYo said:
If player 2 tosses the coin 1 million times, the probability is that they will be first to get 100 points, right?

Edited by JDMFanYo on Tuesday 19th January 14:02
The chance of tossing 1 million heads in a row is infinitesimally small.

If Player 2 tosses it 1 million times and doesn't get 1 million heads they get 0 points.

They will lose 100-0.

Edited by him_over_there on Tuesday 19th January 14:09